AMD drool factor :)

[Soulkeeper]

This has me drooling, gonna be some monsters coming out

 

[Mucker]

Eva Longoria makes me druel ......computer hardware, well, I don't think so.......

 

[Keysplayr]

And it's "drool" guys.

 

[keeleysam]

It's called drooling, and a just a die shrink isn't going to propel AMD over Core... they will have to ramp up clockspeeds.

 

[RichUK]

- HT-3 (upto 5.2GT/sec): Giga transfers, so that looks to be 2.6Ghz speed on the HT link, compared to the current 1Ghz. No real use in 1P systems, but hey.

- Enhanced crossbar, a cross bar that allows more than two cores to reside on.

- DDR2 with migration path to DDR 3: I would have thought this would effect the socket arrangement more than the memory controllers, as those can just be tweaked. Perhaps socket F 1207 provides the extra I/O pins required for DDR3.

- FBDIMM when appropriate: Well, once again pointless on 1P setups. However possible benefits to the server chips being as there is more bandwidth, but a lot more heat a as a by product.

Well this should be the first true quad core on a single die produced by either AMD or Intel, also produced on AMD?s up and coming 65nm. By then I would have thought Intel will be moving onto 45nm fabrication.

I wonder what sort of benefit a more efficient unified L3 cache provides. AMD seems to be taking a totally different approach to memory access compared to Intel, most probably due to Intel still using the FSB model. It seems as if AMD are leagues ahead when comparing the subsystem I/O between the two companies.

 

[Pabster]

Originally posted by: RichUK
I wonder what sort of benefit a more efficient unified L3 cache provides. AMD seems to be taking a totally different approach to memory access compared to Intel, most probably due to Intel still using the FSB model. It seems as if AMD are leagues ahead when comparing the subsystem I/O between the two companies.

That's what we thought, but Conroe trounces all over the X2 despite its lack of an integrated memory controller.

Now, when we move to 4 cores on a single die rather than 2, Intel may have trouble. Time will tell.

 

[996GT2]

Hope this works on AM2 without me having to buy a new board...

 

[thescreensavers]

I still dont understand it

 

[formulav8]

The offical Revision "G" appear to be more than just a shrink. Compare the revision "G" and Revision "F" die photo's and you will seem some actual core work done. Possibly a extra complex decoder and ooo load/store buffers. The 4x Core versions should see a shared L3 Cache. Maybe the DualCore too??

Of course, it will be nothing like what the K8L cores will be getting. But it will be alittle more than just smaller transistors. There could also be some mc work like they did from .13 to .09 venice which gave a decent 2%-6% performance increase. I of course can't see in the future, but there will be changes made to the core on a minor scale that will increase clock per clock performance. But it will NOT be enough to overtake Intels new core. Especially in Integer performance.

Just my penny or 2x is all


Jason

 

[TanisHalfElven]

http://www.google.com/translate?u=http%...ype%3D1&langpair=ja%7Cen&hl=en&ie=UTF8

looks like it will be for am2

 

[formulav8]

Originally posted by: Pabster

Originally posted by: RichUK
I wonder what sort of benefit a more efficient unified L3 cache provides. AMD seems to be taking a totally different approach to memory access compared to Intel, most probably due to Intel still using the FSB model. It seems as if AMD are leagues ahead when comparing the subsystem I/O between the two companies.

That's what we thought, but Conroe trounces all over the X2 despite its lack of an integrated memory controller.

Now, when we move to 4 cores on a single die rather than 2, Intel may have trouble. Time will tell.

Intel did some major homework and did some special stuff on the core to hide/overcome the offchip memory controller. It really has nothing to do with the i/o system.



Jason

 

[Soulkeeper]

you must also take into consideration the doubling of the cache, increasing of the fsb/clock, and ipc all mostly afforded to the conroe from the die shrink
it's amazing how much more can be fit on the die from the shrink to .65
i don't think it is fair to compare a .65 to a .90
this will be amd's answer on a level playing feild imo

 

[coldpower27]

Originally posted by: Soulkeeper
you must also take into consideration the doubling of the cache, increasing of the fsb/clock, and ipc all mostly afforded to the conroe from the die shrink
it's amazing how much more can be fit on the die from the shrink to .65
i don't think it is fair to compare a .65 to a .90
this will be amd's answer on a level playing feild imo

It is fair to compare them when AMD isn't releasing their 65nm products until the end of this year and Intels have been available for sometime.

It won't be too long after 65nm vs 65nm when we will be comparing Intels 45nm to AMD's 65nm.

 

[Baked]

That guy's so funny, put the GPU inside the CPU.

 

[atom]

I don't get it, is there any NEW news on that Chinese site?

From what I can decipher, there isn't. Besides the picture of that ES chip.

 

[Vegitto]

I think you guys are missing something. This is a 65nm, quad-core chip with shared L3 cache. And from what I've heard, it's working out. AMD might have a quad-core out this year.

 

[mayest]

Originally posted by: thescreensavers
I still dont understand it

Why are you guys using a Japanese translator for a Taiwanese site? Here is the Simplified Chinese version.

http://translate.google.com/translate?u...=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

 

[NMDante]

Not to rain on anyone's parade, but it's a picture of a chip.

Unless they can run something with it, it's just a chip. It could be an existing AM2 engineering sample or something, and they are claiming it's the 65nm chip. He tries to explain the chip coding number, but it's still just a chip, working or non-working is unknown.

I have a lot of chips at my house, and I claim they are a never heard of chip being developed in secret, but until I can run something, I'm full of "you know what".

Not saying the guy who started that post is lying, but showing a chip doesn't really inspire me to believe it's the 65nm AMD cpu.

I would like to see that CPU working, but showing a chip is nice, but showing some actual apps running, or even CPU-Z with unknown as the processor would be much better.

 

[lopri]

Originally posted by: RichUK
- HT-3 (upto 5.2GT/sec): Giga transfers, so that looks to be 2.6Ghz speed on the HT link, compared to the current 1Ghz. No real use in 1P systems, but hey.

- Enhanced crossbar, a cross bar that allows more than two cores to reside on.

- DDR2 with migration path to DDR 3: I would have thought this would effect the socket arrangement more than the memory controllers, as those can just be tweaked. Perhaps socket F 1207 provides the extra I/O pins required for DDR3.

- FBDIMM when appropriate: Well, once again pointless on 1P setups. However possible benefits to the server chips being as there is more bandwidth, but a lot more heat a as a by product.

Well this should be the first true quad core on a single die produced by either AMD or Intel, also produced on AMD?s up and coming 65nm. By then I would have thought Intel will be moving onto 45nm fabrication.

I wonder what sort of benefit a more efficient unified L3 cache provides. AMD seems to be taking a totally different approach to memory access compared to Intel, most probably due to Intel still using the FSB model. It seems as if AMD are leagues ahead when comparing the subsystem I/O between the two companies.

Rich, do you read Chinese? :Q
Anyhow thanks for the translation. I thought the 4x4 was a publicity stunt that'd never "practically" materialize. I guess I was wrong..

 

[coldpower27]

Sorry no AMD Quad Core for sale this year. That is slated for Mid 2007 for servers.

 

[Soulkeeper]

Originally posted by: coldpower27
Sorry no AMD Quad Core for sale this year. That is slated for Mid 2007 for servers.

And you know this as fact ?
what sources do you have ?

 

[atom]

Originally posted by: Soulkeeper

Originally posted by: coldpower27
Sorry no AMD Quad Core for sale this year. That is slated for Mid 2007 for servers.

And you know this as fact ?
what sources do you have ?

Because AMD said so on their official roadmap? Plus AMD said they would only "demonstrate" a quad core part by the end of the year. It would be at least several months after demonstration that an actual product rollout would happen if history holds true. So if they were to release some quad core part by the end of the year they would, at the latest, probably would demonstrate a quad core part in the next week or two. And that doesn't look like it's gonna happen.

 

[Vegitto]

What, is your CPU cold with solution?

 

[myocardia]

Originally posted by: Vegitto
What, is your CPU cold with solution?

Huh?

 

[myocardia]

Originally posted by: NMDante
Not to rain on anyone's parade, but it's a picture of a chip.

Actually, if you'll look at that picture closely, it sure does seem to be PhotoShopped.